Accuracy assessment of topography and forest canopy height in complex terrain conditions of Southern China using ICESat-2 and GEDI data

ICESat-2 and GEDI offer unique capabilities for terrain and canopy height retrievals; however, their performance and measurement precision are significantly affected by terrain conditions. Furthermore, differences in data scales complicate direct comparisons of their measurement capabilities. This study evaluates the accuracy of terrain and canopy height retrievals from ICESat-2 and GEDI LiDAR data in complex terrain environments. Jinghong City and Pu'er City in Southwest China were selected as study areas, with high-precision airborne LiDAR data serving as a reference. Ground elevation and canopy height retrieval accuracies were compared before and after scale unification to 30 m × 30 m under varying slope conditions. Results indicate that ICESat-2 shows a significant advantage in terrain height retrieval, with RMSE values of 4.75 m and 4.21 m before and after scale unification, respectively. In comparison, GEDI achieved RMSE values of 4.94 m and 4.96 m. Both systems maintain high accuracy in flat regions, but accuracy declines with increasing slope. For canopy height retrieval, GEDI outperforms ICESat-2. Before scale unification, GEDI achieved an R² of 0.73 with an RMSE of 5.15 m, and after scale unification, an R² of 0.67 with an RMSE of 5.32 m. In contrast, ICESat-2 showed lower performance, with an R² of 0.65 and RMSE of 7.42 m before unification, and an R² of 0.53 with RMSE of 8.29 m after unification. GEDI maintains higher canopy height accuracy across all slope levels. Post-scale unification, both systems show high accuracy in ground elevation retrieval, with ICESat-2 being superior. In contrast, GEDI achieves better canopy height retrieval accuracy. These findings highlight the synergistic strengths of ICESat-2's photon-counting and GEDI's full-waveform LiDAR techniques, demonstrating advancements in satellite laser altimetry for terrain and canopy height retrieval.